Measuring or Exposures

Islamic Economic Studies
Vol. 20, No.1, June 2012
Measuring Operational Risk Exposures in Islamic Banking:

A Proposed Measurement Approach
HYLMUN IZHAR
Abstract
The aim of the paper is to propose a model, namely Delta-Gamma Sensitivity
Analysis-Extreme Value Theory (DGSA-EVT). DGSA-EVT is a model to
measure HF-LS and LF-HS type of operational risks. The first leg of the
proposed model, namely DGSA, is a methodology that deals with
propagation of errors in the value adding activities which works by using
measures of fluctuations in the activities.
The sensitivities of the output, hence, are deployed to estimate the
performance volatility. Furthermore, the second leg of the proposed model,
Extreme Value Theory (EVT), is a technique to cater for an excess
operational loss over a defined threshold
which is normally characterised by low frequency and high severity (LF-HS)
type of loss.
1. Introduction
The complexity of operational risk measurement has been exacerbated by two
major dimensions of operational risk data, namely high frequency-low severity
(HF-LS) and low frequency-high severity (LF-HS)1, and the integration of scaling
external and internal data2. Consequently, each type requires a different approach
The author would like to extend his gratitude to IRTI scholars and the participants of IRTI
Research Seminar for their helpful suggestions and stimulating discussion. He is also
grateful to three anonymous referees whose insightful comments have enabled him to
improve the paper. Any shortcoming therein, however, solely rests upon the author. He can
be reached at hizhar@isdb.org.
1
2
See Moosa, 2007

See Chernobai, Anna S., Rachev, Svetlozar T., and Fabozzi, Frank J., 2007
45
46 Islamic Economic Studies, Vol. 20 No. 1
to cater for operational risk. The current literature on operational risk almost
exclusively focuses on two issues: firstly, the estimation of operational risk loss
processes using extreme value theory or Cox processes3, and secondly, the
application of these estimates to the determination of economic capital4. In the
estimation of economic capital for operational risk, the estimates appear to be quite
large, in fact, at least as large as that necessary to cover market risk5. As evidenced
by the references mentioned earlier that the modelling and estimation of
operational risk is treated identically to market and credit risk, i.e., a loss process is
modelled and estimated. However, this is where the similarity comes to an end.
Unlike market and credit risk, which are external to the bank in their origin,
operational risk is internal to the bank.
An intensive use of Value at Risk (VaR) has also taken place in the
measurement of risk exposure in financial institutions. For a long time, economists
have considered empirical behaviour models of banks where these institutions
maximise some utility criteria under a solvency constraint of VaR type6. Similarly,
other researchers have studied optimal portfolio selection under limited downside
risk as an alternative to traditional mean-variance efficient frontiers7. Moreover,
internal use of VaR by financial institutions has also been addressed in a delegated
risk management framework in order to mitigate agency problems8.
Despite a growing interest in VaR related to credit risk and market risk; there is,
unfortunately, a very limited research in the area of operational risk. Nevertheless,
there has not been any research dealing with the theoretical properties of risk
measures and their consequences on operational risk measurement in Islamic
banking. Islamic Financial Services Board (IFSB) as one of regulatory bodies in
Islamic banking industry in its draft No. 2 on Capital Adequacy Standard9
mentions the definition of operational risk and proposes Basic Indicator Approach
3
See Chavez-Demoulin et al., 2006; Coleman, 2003; de Fontnouvelle et al., 2004; de Fontnouvelle et
al., 2005; Ebnother et al., 2001; Jang, 2004; Moscadelli, 2004; Lindskog and MecNeil, 2003 ; K.
Dutta, J. Perry, 2007.
4
See de Fontnouvelle et al., 2004; de Fontnouvelle et al., 2005, Moscadelli, 2004, and Basel
Committe on Banking Supervision, 2009.
5
See Chernobai, Anna S., Rachev, Svetlozar T., and Fabozzi, Frank J., 2007
6
This issue is discussed in Gollier et al., 1996; Santomero and Babbel, 1996.
7
As exemplified in Roy, 1952; Levy and Sarnat, 1972; Arzac and Bawa, 1977.
8
See Kimball, 1997; Froot and Stein, 1998; Stoughton and Zechner, 1999
9
Islamic Financial Services Board, Capital Adequacy Standards for Institutions (Other than
Insurance Institutions) Offering Only Islamic Financial Services, 2005, p. 17, www.ifsb.org. The two
methods mentioned in IFSB Standards are adopted from Basel Committee on Banking Supervision
Standard on International Convergence of Capital Measurement and Capital Standards: A Revised
Framework, 2006.
Hylmun Izhar: Measuring Operational Risk Exposures in Islamic Banking 47
(BIA) and the Standardised Approach (TSA) as methods to calculate operational

risk capital. The proposed methods are basically meant for the calculation of
capital which needs to be kept aside in order to cater for operational risks. There is,
however, an essential step which is overlooked, namely a method to measure the
magnitude of operational risk exposures.10 The paper, hence, provides a proposed
measurement approach to fill this gap11.
The proposed model, namely Delta-Gamma Sensitivity Analysis-Extreme Value
Theory (DGSA-EVT), is a model to measure HF-LS and LF-HS type of operational
risks. The first leg of the proposed model, namely DGSA, is a methodology that
deals with propagation of errors in the value adding activities which works by
using measures of fluctuations in the activities. The sensitivities of the output,
hence, are deployed to estimate the performance volatility. Through operating loss
distribution that is the result of the entire quantification process, DGSA would help
in generating the level of operational value at risk (OpVaR) of the analysed Islamic
banks. Furthermore, the second leg of the proposed model, Extreme Value Theory
(EVT), is a technique to cater for an excess operational loss over a defined
threshold which is normally characterised by low frequency and high severity (LFHS) type of loss.
The second section of the paper reviews in some more detail the existing
models in operational risk measurement and its classifications. The third section
explains the theoretical background of the proposed model and its features. In the
fourth section, attention is focused on the empirical aspect of the proposed model.
The paper concludes with a fifth section, which includes practical suggestions and
some direction for future research.
2. Review of Operational Risk Modelling
Modelling operational risk ranges from mathematical to statistics-econometrical
approach which is designed to measure the regulatory and economic operational
risk capital. Different models are also designed to study causes and consequences
10
This shortcoming is also reflected in the studies by Khan and Ahmed, 2001; Hassan and Dicle,
2005; Ismail and Sulaiman ,2005; Kahf,2005; Muljawan, 2005; and Sundararajan, 2005.
11
It is expected that the proposed approach will result in an operational risk capital charge that
credibly reflects the operational risk profile of the bank. This is the essence of Operational Risk
Management System (ORMS) which consists of the systems and data used to measure operational
risk in order to estimate the operational risk capital charge. ORMS is a subset of Operational Risk
Management Framework (ORMF). For further discussion, see Basel Committee on Banking
Supervision Consultative Document on Operational Risk-Supervisory Guidelines for the Advanced
Measurement Approach, 2010.
of operational risk. Surely, a constantly changing financial environment has made

modelling of operational risk vital12. Furthermore, operational risk modelling is
also needed to provide bank management with a tool to make a better decision in
carrying out a desirable level of operational risk management. It is also suggested
that the only feasible way to effectively manage operational risk is by identifying
and minimising it, which requires the development of adequate quantification
techniques13. As a matter of fact, quantification of operational risk is a prerequisite
for the formulation of an effective operational risk management and a sound
economic capital framework14.
2.1. Taxonomy of Operational Risk Modelling
The paper broadly classifies modelling in operational risk into three classes; (i)
process approach, (ii) factor approach, and (iii) actuarial approach. It should also
be noted that a selection of which approach to implement may largely depend on
operational risk categories (ORC) which may vary across an institution. As clearly
mentioned in the Basel Committee Document on the Supervisory Guidelines for
the Advanced Measurement Approaches, a banks risk measurement is greatly
influenced by the number of ORCs used within the model15.
Table-1
Approaches in Operational Risk Modelling
Process Approach
Causal models
Bayesian belief networks
Fuzzy logic
Reliability analysis
Connectivity
System Dynamics
Factor Approach
Risk Indicators
Capital Assets Pricing
equivalent models
Predictive models
Actuarial Approach
Empirical loss distributions
Explicit distributions
parameterised using historical
data
Extreme value theory
Source: Smithson and Song (2004)

12
A. Peccia, Using Operational Risk Models to Manage Operational Risk, in Operational Risk:
Regulation, Analysis and Management, ed. C. Alexander, 2003, p. 363. London: Prentice HallFinancial Times.
13
K. Bocker and C. Klupperberg, Operational VAR: A Closed Form Approximation, Risk, 2005,
p.90.
14
K. Fujii, Building Scenarios, in Operational Risk: Practical Approaches to Implementation, ed.
E. Davis, 2005, p. 171, London: Risk Books.
15
See Basel Committee on Banking Supervision Consultative Document on Operational RiskSupervisory Guidelines for the Advanced Measurement Approach, 2010, paragraph 33. A proposed
operational risk categories (ORCs) for an Islamic bank is presented in the Appendix
2.2. Process approach

It is an approach that focuses on the chain of activities that comprise an
operation or transaction (in much the same way that an industrial engineer
examines a manufacturing process by looking at the individual work stations).
Examples of this approach include:
Causal models; it attempts to look at a specific outcome (for example, a
settlement payment) in terms of the probabilistic impact of the activities that are in
the chain (for example, recognition that a payment date has occurred, calculation of
the settlement amount, notification of the counterparty, and paying or receiving).
The success of each activity in the chain might be modelled as a function of inputs.
Reliability analysis; it is used in operational research to measure the impact of
failure of components in complex mechanical/electronic system. However, it is
also widely implemented in operational risk to estimate the hazard rate of arrival of
failure (operational risk event)
Connectivity; which requires the modelling process to develop a connectivity
matrix that can then be used to estimate the likelihood of failure (or potential
losses) for the process as a whole.
Three additional techniques that could be considered process approaches are:
Bayesian belief network, which extends the causal model technique by treating
the initial model as the null hypothesis, and so, as data is collected, the model can
be tested to provide a more accurate picture of the process.
Fuzzy logic is a branch of mathematics that facilitates decision-making when
some of the inputs are vague, or if they are subjective judgements. In a causal
model, fuzzy logic could provide a way to aggregate the subjective drivers of a
process.
System dynamics, which extends the connectivity approach; it is carried out by
making the connections between dynamic activities. This technique requires a
development of the model to simulate the cause-effect interactions among activities
that make up the processes within the firm.
2.3. Factor Approach

A factor approach was initiated as an attempt to identify the significant
determinants of operational risk either at the institutional level or at the level of
an individual business or individual process. The objective is to obtain an equation
that relates the level of operational risk for institution i (or business i or process i)
to a set of factors:
( OperationalRisk )i ( Factor1 ) ( Factor 2 )

The key element of factor approach is the identification of appropriate factors in
order to obtain the measures of the parameters , ,and . As a result, an
estimation of the level of operational risk that will exist in future periods can be
materialised. In the analysis of operational risk quantification, Smith and Song
(2004) describe three applications of the factor approach:
Risk indicators, in which regression techniques are utilised to identify the
significant operational risk factors.
Capital Assets Pricing equivalent model; a model that relates the volatility in
share returns (and earnings and other components of the institutions valuation) to
operational risk factors.
Predictive models, which use discriminated analysis and similar techniques to
identify factors that lead to operational losses.
2.4. Actuarial Approach
An actuarial approach attempts to identify the loss distribution associated with
operational risk either at the level of an institution or at the level of a business or
process.
Empirical loss distribution, is the most straightforward way to estimate the loss
distribution, using the institutions own data on losses or both internal data and
(properly scaled) external data. However, empirical loss distributions will probably
suffer from limited data points (especially in the tail of the distribution).
Explicit distributions parameterized using historical data is one way to get
around the sparse data problem. The analyst specifies a distributional form for the
loss distribution or a distribution for the frequency of occurrence of losses and a

different distribution for the severity of the losses.
Extreme value theory provides another way of getting around the data
sparseness problem. This theory is an area of statistics concerned with modelling
the limiting behaviour of sample extremes, which indicates that, for a large class of
distributions, losses in excess of a high enough threshold all follow the same
distribution (a generalised Pareto distribution).
3. Empirical Research in Islamic Banking
In the IFSB Draft No. 2 on Capital Adequacy Standard, operational risk is
defined as the risk of losses resulting from inadequate or failed internal risk and
Shar ah compliance risk (IFSB, 2005: 22)16. This definition is rather different from
Basel 2 on operational risk17. However, IFSB adopts Basel 2s methodology in the
calculation of a minimum capital requirement for operational risk exposure. Four
methods have been proposed by the Basel; namely the Basic Indicator approach
(BIA), the Standardised approach (TSA), the Alternative Standardised approach
(ASA) and the Advanced Measurement approach (AMA). BIA takes the moving
average of gross income as a proxy of the size of operational risk exposure and
suggests a parameter of 15% to calculate the minimum capital required to stand for
this kind of risk. TSA is a little more refined as it takes average gross income at the
activity level after dividing a banks activities into 8 categories and suggests a
parameter for each of them ranging between 12 and 18 percent. Under the ASA, for
retail and commercial banking business lines, loans and advances replace gross
income as the proxy indicator. Finally, AMA allows using internal measurement
methodologies to calculate the minimum capital requirement for operational risk
exposure provided the bank satisfies certain qualification criteria to assure the
supervisory authority of the existence of efficient and independent operational risk
management system and of its ability to fairly estimate operational risk and the
capital needed to face it, including the expected losses as well as the unexpected
losses.
The IFSB standards provide fairly detailed guidance on adaptation of Basel 2 to
the specific risk characteristics of Islamic banks. In particular, the IFSB draft
proposes an adaptation of standardised approach to operational risk
16
Islamic Financial Services Board, op cit. p. 22.

In Basel 2, operational risk as the risk of loss resulting from inadequate or failed internal processes,
people and systems or from external events. This definition includes legal risk, but excludes strategic
and reputational risk. See Basel 2, 2005, paragraph 33.
17
measurementbased on externally provided rating categoriesand within this

framework allows supervisory discretion to recognise the extent of risks assumed
by the PSIAs18 in computing capital adequacy for Islamic banks. Kahf opposes the
use of gross income as a proxy of operational risk exposure as set out by IFSB19. In
this respect, his argument is in line with Sundararajan who argued that the use of
gross income as the basic indicator for operational risk measurement could be
misleading in Islamic Banks, as large volume of transactions in commodities, and
the use of structure finance raise operational exposures that will not be captured by
gross income.20 However, Sundararajan (2005) still supports the standardised
approach that allows for different business lines to be better suited, but would still
need adaptation to the needs of Islamic banks.
Some empirical aspects of the operational soundness in Islamic banks were
studied by Ismail and Suleiman (2005), Hassan and Dicle (2005) and Muljawan
(2005). Using the Cavello and Majnoni model, Ismail and Suleiman discuss the
interaction between the capital requirement as stated in the New Basel Capital
Accord and the cyclical pattern of profit21. In addition to that, CAMEL22
framework is deployed by Muljawan as an alternative tool to assess the operational
soundness of Islamic banks.23 The analysis of Hassan and Dicle is somewhat
broader than other papers in the sense that it also discusses the nature of
18
PSIA refers to profit sharing investment account

Monzer Kahf, Basel II: Implications for Islamic Banking. Paper presented at the Sixth
International Conference on Islamic Economics and Finance: Islamic Economics and Banking in the
21st Century, organized by Bank Indonesia, Islamic Research and Training Institute, The
International Association for Islamic Economics, and University of Indonesia, 2005, p. 313
20
V. Sundararajan, Risk Measurement and Disclosure in Islamic Finance and the Implications of
Profit Sharing Investment Accounts. Paper presented at the Sixth International Conference on Islamic
Economics and Finance: Islamic Economics and Banking in the 21st Century, organized by Bank
Indonesia, Islamic Research and Training Institute, The International Association for Islamic
Economics, and University of Indonesia, 2005, p. 127.
21
Abd. Ghaffar Ismail and Ahmad Azam b. Sulaiman, Cyclical Patterns in Profits, Provisioning and
Lending of Islamic Banks and Procyclicality of the New Basel Capital Requirement. Paper presented
at the Sixth International Conference on Islamic Economics and Finance: Islamic Economics and
Banking in the 21st Century, organized by Bank Indonesia, Islamic Research and Training Institute,
The International Association for Islamic Economics, and University of Indonesia, 2005, p. 280.
22
C(apital), A(sset quality), M(anagement), E(arning), and L(iquidity)
23
Dadang Muljawan,, A Design for Islamic Banking Rating System: An Integrated Approach. Paper
presented at the Sixth International Conference on Islamic Economics and Finance: Islamic
Economics and Banking in the 21st Century, organized by Bank Indonesia, Islamic Research and
Training Institute, The International Association for Islamic Economics, and University of Indonesia,
2005, p.317
19
operational risk.24 However, it does not make any suggestions on how to handle
capital requirements with respect to Islamic banks.
There are two things that can be highlighted from the review above; first, it is
clear that the empirical research that have been conducted are on the aspect of
capital attribution for operational risk; second, there is no unanimous standard of
operational risk measurement method. The most recent research on this issue was
conducted by Jackson-Moore; nevertheless, the writer could not come up with a
conclusive suggestion on the refined measurement method.25
The following section attempts to discuss the proposed framework in measuring
operational risk exposures in Islamic banks.
4. Delta-Gamma Sensitivity Analysis (DGSA): A Proposed Approach
The objective of operational risk management is to decide which risks are
important to the bank so that it could determine their magnitude and mitigate them
accordingly. Therefore a refined measurement method is required to provide a
measure that has a defined relationship to a risk factor that can be assigned as
controllable or uncontrollable. This would result in the determination of an
appropriate intervention for controllable risks by focusing on their causes.
Given the foregoing discussion, the impact on operations can be separated into
controllable risk and uncontrollable risk. In this study, a controllable risk is defined
as any risk which has assignable causes that can be influenced. Generally, processrelated risks will have assignable causes and therefore, they are controllable. For
instance, classifying loan customers into the wrong credit categories can result in
substantial differences in the default rates and loan provision requirements and is
an example of a risk that is controllable because the cause is known.
Uncontrollable risk, on the other hand, is defined as any risk that does not have
causal factors that can be influenced. Their impact is determined through loss
24
M. Kabir Hassan and Mehmet F. Dicle, Basel II and Capital Requirements for Islamic Banks. Paper
presented at the 6th International Conference at Islamic Economics, Banking and Finance, organised
by Islamic Research and Training Institute-Islamic Development Bank, Bank Indonesia, and Ministry
of Finance Republic of Indonesia, 2005, p. 255
25
Elisabeth Jackson-Moore, Measuring Operational Risk, in Islamic Finance: The Regulatory
Challenge, eds. Simon Archer and R.A.A.Karim, 2007, p. 239, Chichester: John Wiley & Sons (Asia)
Pte Ltd.
models that analyse extreme values (losses), and use classification instead of
causes. Ideally, extreme loss models will be used with scenarios that provide stress
points for the analysis. Uncontrollable does not mean that there is nothing that can
be done about it. There are many mitigation strategies that can be implemented in
order to reduce the effects of a loss. Also, uncontrollable risks may become
controllable if an assignable cause can be found and which would enable the
management to carry out a corrective action.
The proposed DGSA deals with controllable risks; in other words, DGSA is
designed to measure the magnitude of operational risk exposures which can be
controlled, or HF-LS type of operational risks.
4.1. Building Blocks of DGSA
The analysis of DGSA begins by developing a function for a value adding
process and then examining the key factors that contribute to the performance and
their associated errors (uncertainties). This can be done by partitioning the business
unit into different income generating channels (IGCs). IGCs contain different
earnings function as the unit of analysis for measuring operational risk, as shown in
Figure 1.
Income generating channels can be defined as the production unit by which a
bank creates a product valuable to its customers. An activity in the income
generating channels employs purchased inputs, human resources, capital, and some
form of technology to perform its function26. Since a business unit has profit and
loss reporting (by definition), its income generating processes are the key
components that make up the profit and loss for the business unit. In our model,
Islamic banks business model can be partitioned into three income generating
channels, namely; (a) investment channel, (b) financing channel, and (c) service
channel.
a) Investment channel, which comprises any investment in the form of a
partnership. There are two types of investing instruments: fund management
(mu rabah) and equity partnership (mush rakah). Mu rabah, which can
be short, medium, or long term, is a trust-based financing agreement
whereby an investor entrusts capital to an agent to undertake a project.
Profits are based on a pre-agreed ratio. Mush rakah, which can be either
26
Jack L. King, Operational Risk: Measurement and Modelling, 2001, p. 74, Chichester: John Wiley
& Sons, Ltd.
medium or long term, is a hybrid of shirka (partnership) and mu rabah,

combining the act of investment and management. In the absence of debt
security, the Shar ah encourages this form of financing.
b) Financing channel, which contains any financing instruments that are used
primarily to finance obligations arising from the trade and sale of
commodities or property. Financing instruments also include instruments
generating rental cash flows against exchange of rights to use the assets such
as ij rah and isti n . Financing instruments are closely linked to a sale
contract and therefore are collateralised by the product being financed. These
instruments are the basis of short-term assets for the Islamic banks.
Mur ba ah, a cost-plus sales contract, is one of the most popular contracts
for purchasing commodities and other products on credit.
c) Service channel; consists of any financial transactions that create earnings by
charging fees, an example of which is ju lah.
For each income generating channel, an earning figure can be located and
linked up with causal factors for the business. Causal factors can be defined as
factors that have impacts on earnings. In other words, DGSA uses risk factors
resulting from causal factors that create losses with a random uncertainty to
measure the variability of earnings.
In contrast, un-assignable loss cannot be tied to a risk factor since the cause is
normally unknown or is due to an external event. Based on the causality between
risk factors contributing to assignable losses, an earning function can be produced
in each income generating channel. The DGSA methods use factors which lead to
loss and their sensitivities to generate loss distributions in different business units.
Figure-1
How Does the DGSA Work?
High Frequency-Low Severity (HF-LS):
predictable, assignable, and controllable
I.
Identification of Risk Factors in

Three Income Generating Channels (IGC)
Investment
II.
Services
Financing
Establish earnings functions related to risk factors in each IGC. How?

Delta-Gamma based on Error Propagation
1st step (static process): Given

E=f(x); sensitivity of causality
factors is defined as
E f
The Delta
x x
2st step (dynamic process):

the sensitivity of delta is defined as
E 2 f
The Gamma
x x 2
Output:
Operating Loss Distribution (OLD)
Value at Risk for

Operational Losses
Decide maximum OLD as

a threshold ()
In it worth noting here that losses within business units are not normally
accounted for in a systematic way that would allow their direct assignment to risk
factors. Since there are a large number of small losses, many banks simply
aggregate operational losses in general accounts along with other entries. They may
be included as a cost of doing business or simply mixed up in the profit and loss
accounting. Without having a loss figure that can be linked to risk factors,
therefore, it is almost impossible to produce a direct measurement of operational
risk caused by assignable loss. Hence the DGSA method can overcome this
problem.
In summary, the steps of building DGSA frameworks are as follows:
1)
2)
3)
4)
5)
6)
Establish the business model with income generating channel

Determine the risk factors for the major activities in the income generating
channel
Determine the relations between risk factors and earning through setting
up earnings function in different income generating channel
Estimate operational losses using uncertainty of the risk factors propagated
to the risk in earnings (Delta-Gamma method)
Set the threshold of operating losses from the processes using the risk
factor uncertainties and operating losses from Delta-Gamma method
Filter the large losses using the threshold.
4.2. Key Features of DGSA

The DGSA methodology is the calculation technique to determine the value of
the assignable losses based on the sensitivity causality between the risk factors.
DGSA is produced through error propagation of the risk factors to measure
operational risk. The uncertainty of the risk factors is utilised to calculate the
uncertainty in earnings using sensitivities from which the relation of the change in
earnings to a change in the risk factors can be located.
In DGSA, operational risk is measured as the uncertainty in earnings due to two
parts. First, using the uncertainty in causal factors for losses up to a threshold and
second, using a large loss model for un-assignable loss above a threshold. Causality
model, hence, plays a critical role in determining the risk factors establishing the
model. Hence, the combination of the two constitutes DGSA and is described by
the operational risk formula as follows:
u( E ) fu( X 1 )...u( X n )) ( Lunassignable Lunassignable )
(1)
Uncertainty in earnings due to operational risk is a function of the uncertainties in a
set of risk factors plus a function of the distribution of un-assignable losses larger
than a given threshold (). DGSA method is used to calculate the first term in the
model. This model expresses the uncertainty in earnings as a function of the
uncertainty in a set of risk factors:
u( E ) f ( u( X1 )...u( X n ))
(2)
DGSA method for measuring operational risk is based on the five following key
concepts:
1. Earnings as a function of causal factors.
In DGSA method, it is assumed that earnings are described by a series of causal
factors. For a given earnings level, there is a set of causal factors whose values
are used to estimate earnings:
earnings = f (causal factors)
(3)
Earnings are described as a function of a set of causal factors. For example,

earnings may be calculated as 20% of sales revenue minus an adjustment for
rejects. By separating the causal factors into constants and volatilities, earnings
can be described by a set of performance drivers that create the expected level
of earnings and a set of risk factors that create volatility in the level of earnings
(risk):
earnings = f ( performance drivers) f (risk factors)
(4)
Earnings are described as a function of performance drivers for level and risk
factors for volatility. Therefore, in the model, earnings are calculated as 20% of
sales revenue minus the variance to target cost for rejects. Sales revenue is
the performance driver and rejects is the risk.
2. The risk in earnings is a random fluctuation in value caused by uncertainty in
the risk factors. Given
(5)
E f(x)
Therefore
u( E ) f ( u( x ))
(6)
3. The basic measure of uncertainty for operational risk is the standard deviation
of the mean, or standard error.
In general, the standard deviation of the mean of the measured values is referred to
as the standard error or simply the error27. It is calculated from a sample of n
measures using the following formula:
n
1
( xk x )2
n( n 1 ) k 1
(7)
Whereby x is the mean of the analysed operational risk variable.

4. Uncertainties are combined using the formula for the expected value of the sum
of variances.
This formula is given for the simple case of correlation values of only 0 or 1,
corresponding to independent analysed operational risk variables and others
that are perfectly correlated. Normally this should be sufficient for operational
risk measures.
j
i
j
2
z
2
i
(8)
Formula for combining uncertainties using standard errors where the is are
uncorrelated and the js are correlated (perfectly) measures.
5. Uncertainties for functions of uncertainty measures are calculated using the law
of error propagation. For each risk factor, the sensitivity of the earnings with
respect to the factor is needed. The sensitivity is the amount of change in earnings
given a single unit change in the factor with everything else remaining unchanged,
or the partial derivative of the earnings function with respect to the factor. Given
the earnings function that expresses earnings as a function of a factor
E f(x)
27
Aswath Damodaran, Corporate Finance: Theory and Practice, 2001, p. 153, New York: John
Wiley & Sons, Inc.
Then sensitivity is defined as

E f
x x
(9)
The method of combining measurement uncertainties from various factors and

accounting for their correlation is known as the propagation of uncertainty. The
basic formula uses the sensitivities (partial derivatives) of the factors to calculate
the standard deviation of the estimate. It is based on a Taylor approximation for the
uncertainty in terms of factors such as:
R f ( X 1, X 2 ...X n )
(10)
Using the Taylor approximations first term, the uncertainty for the measure can be
figured out using the following technique:
n
f
u 2 ( r )
i 1 x i
N 1 N
2
f f
u ( xi ) 2
i j ij
i 1 j i 1wi w j
(11)
The formula shown above is the formula for the calculation of combined
uncertainty from many factors, also known as the general law of error
propagation. Where u( ) denotes the uncertainty in the value, r is the risk
measurement, x is the factor, and f is the functional relationship between x and r.
The partial derivative term is known as the sensitivity to the factor. This formula
also explicitly considers correlation between factors ij .
6. The gamma ( ) of a portfolio on an underlying assets is the rate of change of
the portfolios delta with respect to the price of the underlying asset. While the
delta is the first derivative of the model, the gamma is the second partial derivative
of the portfolio with respect to different risk factors:
2
Gamma
S 2
(12)
If the value of gamma is small, the delta changes slowly and adjustments to keep a
portfolio delta neutral only need to be made relatively infrequently. However, if
gamma is large in absolute terms, then delta is highly sensitive to the price of the
underlying asset. It is then quite risky to leave a delta neutral portfolio unchanged
for any length of time. In this study, gamma is an important factor in determining
which risk factors are more influential to income generating channels.
It is expected that partnership type of financing, such as mu rabah and

mush rakah would give higher value since they are likely to increase the level of
operational risk exposures.
7. Threshold; it is used to separate losses to be analysed using DGSA from those
that are not assignable. As highlighted in the earlier paragraph, DGSA deals with
small losses (HF-LS type of operational risks); hence, the threshold is the transition
point from small loss (HF-LS) to large loss (LF-HS). However, to ensure that there
will not be any overlap, meticulous calculations must be carried out to set the
threshold precisely since losses assigned to risk factors using DGSA method are
assumed to have random error properties. And DGSA is used to estimate the central
tendency of this uncertainty.
4.3. Why Sensitivity Analysis?
The activity in the field of sensitivity analysis (SA) has been steadily growing, due
to the increasing complexity of numerical models, whereby SA has acquired a key
role in testing the correctness and corroborating the robustness of models in several
disciplines. This has led to the development and application of several new SA
techniques28. Most of the recent literature in portfolio management has proposed
SA approaches based on partial derivatives (PDs)29.
Nevertheless, recent studies in the SA literature have highlighted that PDs-based SA
suffers from several limitations when used for parameter impact evaluation and
risk management purposes30. It is shown that a PDs-based SA to evaluate the
impact of parameter changes with respect to the generic model output31:
1) is equivalent to neglecting the relative parameter changes, or, equivalently,
to impose that all the parameters are varied in the same way;
2) does not allow the appreciation of the model sensitivity to changes in groups
of parameters
28
Studies by Borgonovo and Apostolakis, 2001a; Saltelli, 1997; Saltelli, 1999; Saltelli, Tarantolla and
Chan, 1999; Turany and Rabitz, 2000 show the importance of sensitivity analysis approach in
financial analysis.
29
As brought forward by Drudi, Generale and Majnoni, 1997; Gourieroux, Laurent and Scaillet,
2000; Manganelli, 2004; McNeal and Frey, 2000.
30
See Borgonovo and Apostolakis, 2001a; Borgonovo and Apostolakis, 2001b; Borgonovo and
Peccati, 2004; Borgonovo and Peccati, 2005; Cheok, Parry and Sherry, 1998.
31
As argued by Borgonovo and Apostolakis, 2001a; Borgonovo and Apostolakis, 2001b; Borgonovo
and Peccati, 2004.
Therefore, using Elasticity (E) is considered to be a better alternative as

compared to PDs32. In this case limitation 2 would still be in place, as E is not
additive; and limitation 1 would be replaced by introducing E to impose on any
parameters that are changing by the same proportion.33
This study will show that the use of Differential Importance Measure (D) would
overcome the two above mentioned limitations.
Let us consider the generic model output:
Y = f(x)
(13)
Where x = {xi, i = 1,2,,n} is the set of the input parameters. Suppose:
dx = [dx1, dx2,,dxn]T
Which denote the vector of change; if f(x) is differentiable, then the differential
importance of xs at x0 is defined as34
Ds ( x 0 , dx )
f xs ( x 0 )dx s
df s ( x 0 )
df ( x 0 ) nj1 f j ( x 0 )dx j
(14)
D can be interpreted as the ratio of the (infinitesimal) change in Y caused by a

change in xs and the total change in Y caused by a change in all the parameters.
Thus, D is the normalised change in Y provoked by a change in parameter xs. It can
be shown that35:
a) D shares the additivity property with respect to the various inputs, for
example, the impact of the change in some set of parameters coincides with
the sum of the individual parameter impacts. More formally, let S {1,2,,
n} identify some subset of interest of the input set; hence it would give:
32
CP. Simon, C.P. and L. Blume, Mathematics for Economists, 1994, p. 127. New York: W.W.
Norton & Co
33
See (Borgonovo and Apostolakis, 2001a; Borgonovo and Apostolakis, 2001b; Borgonovo and
Peccati, 2004; Borgonovo and Peccati, 2005).
34
E. Borgonovo and L. Peccati, Sensitivity Analysis in Investment Project Evaluation,
International Journal of Production Economics, 90 (2004), p. 20
35
As concluded by Borgonovo and Apostolakis, 2001a; Borgonovo and Apostolakis, 2001b;
Borgonovo and Peccati, 2004; Borgonovo and Peccati, 2005.
Ds ( x 0 ,d x )
0
sS f s ( x )dxs
Ds ( x 0 ,dx )
0
nj 1 fj( x )dx j sS
(15)
As a consequence,
n
0
Ds ( x ,dx ) 1
(16)
s 1
for example, the sum of the Di (i=1,,n) of all parameters is equal to

unity.
b) Equation (2) shows that D accounts for the relative parameters changes
through the dependence on dx. In fact, equation (14) can be rewritten as:
f s ( x0 )
Ds ( x 0 ,dx )
0
nj 1 f j ( x )
(17)
dx j
dxs
In the hypothesis of uniform parameter changes (H1) (dxj=dxs j,s), the

following can be produced:
D1s ( x 0 )
f s ( x0 )
0
nj 1 fj ( x )
(18)
Hypothesis of proportional changes (H2) dx j j , would result in:

0
xj
D2s ( x0 )
fs( x ) x
0
0
nj 1 fj( x ) x j
0
0
s
(19)
It can be shown that D generalises other local SA techniques as the Fussel-Vesely

importance measure and Local Importance Measure based on normalised partial
derivatives, also known as Criticality Importance or E. More specifically, in case
H2 it holds that36:
D2s ( x 0 )
E ( x0 )
0
Ej( x )
s
n
j 1
(20)
where Es(x0) is the elasticity of Y with respect to xs at x0. Equation (20) shows that
E produces the importance of parameters for proportional changes.
36
E. Borgonovo and L. Peccatti, op cit., p. 23
5. Determination of Risk Factor Contribution

From the practitioners viewpoint, a pertinent issue is how much each of the
process contributes to the risk exposure37. If it turns out that only a fraction of all
processes significantly contribute to the risk exposure, then the risk manager
should only focus on this particular process. It is, therefore, important to analyse
how much each single process contributes to the total risk. This study considers
operational value at risk (OpVaR) resulting from operating loss distribution as a
risk measure. To split up the risk into its process components, this study compares
the incremental risk (IR) of the processes.
Let IR (i) be the risk contribution of process i to OpVaR at the confidence level .
IR (i) = OpVaR (P) -OpVaR(P\{i}),
(21)
Where P\{i} is the whole set of workflows without process i. Since the sum over all
IRs is generally equal to the OpVaR, the relative incremental risk (RIC(i)) of
process i is defined as the IR(i) normalised by the sum overall IR, i.e.
RIC ( i )
IR ( i ) OpVaR ( P ) OpVaR ( P \ { i })
IR ( j )
IR ( j )
j
(22)
as a further step, for each , this paper counts the number of processes that exceed
a relative incremental risk of 1%. The resulting curve is attributed as parameter or
the Risk Selection Curve (RiSC)
6. Extreme Value Theory (EVT)
Extreme value theory (EVT) is a field of study in statistics that focuses on the
properties and behaviour of extreme events. In general, there are two main kinds of
model for extreme values. The most traditional models are the block maxima
models; these are models for largest observations collected from large samples of
identically distributed observations. The second type of model which is more
comprehensive is the peak over threshold (POT) model; this is a model for all large
observations that exceed some high level, and is generally considered to be the
most useful for practical applications, due to their more efficient use of the data
(often limited) on extreme outcomes.
37
S. Ebnother et al., Modelling Operational Risk, 2001, p. 5, Working Paper, ETH Zurich
In our analysis, the application of EVT as the second leg of the proposed model
starts after the determination of a transition point resulting from DGSA. It is
important to note that the transition point is typically classified as the maximum
threshold. EVT offers a parametric statistical approach for the extreme values of
data. Its roots are in the physical sciences and it has recently been applied to
insurance. Since traditional statistical techniques focus on measures of central
tendency (e.g. mean), they are not as accurate when estimating values very far from
the centre of the data. EVT, on the other hand, deals only with the extreme values
and ignores the majority of the underlying data and its measures in order to provide
better estimates of the tails.
The EVT methodology for operational risk is basically a loss model for large
losses using a GPD for the severity. The technique for fitting the GPD to data is the
peaks over threshold method (POT), where large values over a specific threshold
are fitted to the GPD. Following Chavez-Demoulin et al.38, the POT method
deployed in the analysis uses the following basic assumptions:
The excesses of an independent identically distributed (or stationary)
sequence over a high threshold u occur at the times of a Poisson process;
The corresponding excesses over u are independent and have a GPD;
Excesses and exceedance times are independent of each other.
6.1. Operating Framework for EVT
As depicted in Figure 2, the steps for operating EVT in our analysis start with
the separation of loss amount into its severity and frequency.
Furthermore, excess losses are fit to a GPD to determine the severity of a loss
given that it exceeds the threshold. This is a conditional severity distribution for
large losses. Since the number of exceedances follows a Poisson distribution, it is
fitted and used to estimate the frequency of exceedances. Combining the severity
and frequency distributions in a Monte Carlo simulation gives the excess loss
distribution. The resulting excess loss distribution is a multi-period loss distribution
for only those losses that exceed the threshold.
38
V. Chavez-Demoulin, P. Embrechts, J. Neslehova, Quantitative Models for Operational Risk :

Extremes, dependence and aggregation, Journal of Banking & Finance (30), 2006, p. 2639
Figure-2
The Application of Extreme Value Theory for
Operational Risk Measurement in Islamic Banks
Low Frequency-High Severity (LF-HS):
unpredictable, un-assignable, and uncontrollable
Severity [ loss amount>threshold ()]
Frequency distribution
Maximum likelihood method
Fitted to GPD (Generalised Pareto Distribution)
Fitted to Poisson Distribution
Combined in Monte Carlo Simulation

Output:
Excess Loss Distribution (ELD)
Value at Risk for

Operational Losses
6.2. Theoretical Building Blocks of EVT: Fisher-Tippet-Gnedenko Theorem

The Fisher-Tippet-Gnedenko theorem states that given a sample of independent
identically distributed loss data x1,x2,, xn, as the number of observations n
becomes increasingly large, the maximum of the sequence of observations, under
vert general conditions, is approximately distributed as the generalised extreme
value (GEV) distribution with cumulative probability distribution function
1

x
exp
for 0
1
F x
x
exp exp for 0

(23)
where is the location parameter, > 0 is a scale parameter, 1 + z > 0,

, 0, and is the tail index parameter. The GEV has three forms; if
>0, then the distribution takes the form of a type II (Frechet) heavy-tailed
distribution. For <0, the distribution is takes the type III (Weibull) distribution.
When =0, the distribution is the type I (Gumbel) light-tailed distribution. In fact,
the larger the tail index parameter, the fatter is the tail.
6.3. Parameter Estimation
The parameter and can be estimated from the sample mean and sample
standard deviation, respectively. If we rank the data in order size so that
x1>x2>>xn, the tail index parameter can be estimated using the Hill estimator:
k 1
Method I: k 1 ln x j ln xk
k 1 j 1
(24)
Method II: k 1 ln x j ln xk
k j 1
(25)
The problem now is how to choose k. Theory gives little advices as to what value
to choose. Furthermore, the actual estimate will be sensitive to the value of k
chosen. In practice, the average estimator, using either of the following two
formulas, often works well:
n
k 1
Method 1: 1 i where k 1 ln x j lnxk for k =1, 2, n
n i 1
k 1 j 1
(26)
1n
n i 1
1
k
Method 2: i where k ln x j ln xk for k = 1, 2 , n

k
j 1
(27)
6.4. Severity Model

An alternative EVT approach to calculate OpVaR is to use peaks over threshold
(POT) modelling. The underlying principle of the operating framework is to use
peaks over threshold. Although the method of block maxima utilises the FisherTippet-Gnedenko theorem to inform us what the distribution of the maximum loss
is, POT uses the Picklands-Dalkema-de Hann to inform us what is the probability
distribution of all events greater than some large present threshold. The PicklandsDalkema-de Hann theorem states that if Fu is the conditional excess distribution
function of values of the ordered losses X above some threshold, is given by
Fu Pr obX y X ,0 y x F . Then for a suitably high threshold the
limiting distribution of Fu is a generalised Pareto distribution (GPD) with
cumulative distribution function
GPD
1 1 x i , 0
1 exp x , 0
(28)
where > 0, and x 0 when 0 and 0 x -/ when < 0. The parameters

and are referred to, respectively, as the shape and scale parameters. In other
words, ys are called excesses whereas xs are called exceedances.
It is possible to determine the conditional distribution function of the excesses (i.e,
ys) as a function of x:
Fu y PX u y X u
Fx x Fx u
1 Fx u
(29)
In this representations the parameters is crucial, when = 0, we have an

exponential distribution; when < 0, we have a Pareto distributionII Type and
when > 0, we have Pareto distributionI Type. Moreover, this parameter has a
direct connection with the existence of finite moments of the losses distributions.
We have the following equations:
E x k if k 1 /
(30)
Hence in the case of a GPD as a ParetoI Type, when 1, we have infinite

mean models, as also shown by Moscadelli39 and Neslehova et. al40.
Following Di Clemente-Romano41, we suggest to model the loss severity using the
lognormal for the body of the distribution and EVT for the tail in the following
way:
ln x i
0 x u i
i
Fi x
1
i
1 N u i 1 i x u i
u i x
Ni
i
(31)
Where
= standardised normal cumulative distribution functions
Nu(i) = number of losses exceeding the threshold u(i)
N(i) = number of loss data observed in the ith ET
(i) = scale parameters of a GDP
(i) = shape parameters of a GDP
An important issue to consider is the estimation of the severity distribution
parameters. While the estimation maximum likelihood (ML) in the lognormal case
is straightforward, in the EVT case, it is extremely important to consider whether
ML or the alternative probability weighted moment (PWM) routines are able to
capture the dynamics underlying losses severities.
With respect to ML, the log-likelihood function equals
1 n
l , ; X n ln 1 ln1 X i
i 1
(32)
This method works well if > -1/2. In this case, it is possible to show that
39
M. Moscadelli, The Modeling of Operational Risk: Experiences with the Analysis of the data
collected by the Basel Committee, Working Paper, Temi di Discussione del Servizio Studi, No. 157,
Banca dItalia, Roma, 2004.
40
J. Neslehova, P. Embrechts, and V. Chavez-Demoulin, Infinite Mean Models and the LDA for
Operational Risk, Journal of Operational Risk, 1 (1), 2006.
41
A. DiClemente and C. Romano, A copula-extreme value theory approach for modelling
operational risk, in Operational Risk Modelling and Analysis: Theory and Practice, ed. M.G. Cruz.
London: Risk Books, 2004.
n1 2 n , n 1
N 0, M 1 ,n

(33)
1
M 1 1
1
(34)
where
Instead, the PWM consist of equating model moments based on a certain

parametric distribution function to the corresponding empirical moments based on
the data. Estimated based on PWM are often considered to be superior to standard
moment-based estimates. In our case, this approach is based on these quantities:
wr E Z GP D , Z
r
r 1r 1
r 0,1,...
where GP D , 1 GPD , , Z follows a GPD
(35)
(36)
From the above equations, it is possible to show that

2w0 w1 and
w0
2
w0 2w1
w0 2w1
(37)
Hosking and Wallis42 show that PWM is a viable alternative to ML when 0. In

our analysis, we estimated the GPD parameters using the previous approaches
together with the standard Hill estimator.
6.5. Frequency Model
Having fitted a GPD to the amount of loss for a set of excess losses, the next
step is to determine the frequency of losses using a Poisson distribution. The
Poisson distribution is well known as a single parameter distribution for the
number of occurrences of an event with relatively small probabilities given a
relatively large sample. The formula for the Poisson distribution is
e x
Pr( x )
42
x!
(38)
J.R.M. Hosking and J.R. Wallis, Parameter and Quantile Estimation for the Generalized Pareto
Distribution, Technometrics, 29 (3), 1987, p. 344.
Formula for Poisson distribution of x events with single parameter , the arrival
rate. The fitting of the Poisson to a set of occurrences proceeds using the interarrival times for the loss events. That is, the average time between events can be
used to determine the arrival rate or lambda for the Poisson formula. (The arrival
rate is simply the inverse of the inter-arrival time). For the Poisson distribution, it
can be shown that the maximum likelihood estimator for is given by the mean
arrival rate formula below
knk
k
n
(39)
Formula for estimating for the Poisson distribution; where
k is the number of events in a period,
nk is the number of periods with k events,
n is the total number of periods.
A goodness of fit statistic for the Poisson distribution can be found using a simple
2-squared test. The test statistic is:
( n n Pr( k ; ))2
2 k
k
n Pr( k ; )
(40)
Chi-squared test statistic for the goodness of fit of the Poisson distribution to a set
of data; where Pr(k; ) is the probability of k events for the Poisson distribution
with parameter . The degrees of freedom are n-2.
6.6. Compounding via Monte Carlo Methods
Once severity and frequency distributions have been estimated, it is necessary to
compound them via Monte Carlo methods to get a new data series of aggregate
losses, so that we can then compute the desired risk measures, such as the VaR and
expected shortfall.
The random sum L=X1+ + Xn (where N follows a Poisson distribution) have
distribution function:
FL (x) = Pr (L x)
=
p n Pr L x N n
n 0
= pn Fx* n x
n0
(41)
where Fx(x) = Pr ( X x) = distribution function of the severities Xi

Fx* n = n-fold convolution of the cumulative distribution function of X.
Hence, the aggregation of frequencies and severities is performed as a sum of
severities distribution function convolutions, thus determining a compound
distribution, whose density function can be obtained by:
f L x pn Fx* n x
n0
(42)
This aggregation is computed via convolution using Monte Carlo methods. It

should also be noted that the convolution is a bit more complex as the severity
distribution is split in two parts: the body of the distribution, which follows a
lognormal distribution, and the tail, which follows a GPD. As a result, two
different severity levels are generated. Hence, the probability associated at each
severity (i.e., the number of observations obtained by the Poisson distribution) has
to be congruent with the fact that losses may belong to the body or to the tail.
Therefore, it is crucial to consider F(u), where u is the GPD threshold and F is the
distribution function associated to this point. After having sampled from the two
severity distributions, every single loss Xi whose F(Xi) < F(u) will be modeled as a
lognormal variable, otherwise it will be a GPD random draw.
As shown in Figure 1 and Figure 2, value at risk is generated from both
processes, DGSA and EVT. As a result, an approximation of the magnitude of
operational risk is generated by adding the value at risk resulting from DGSA and
EVT processes.
7. Concluding Remarks
Indeed, quantifying operational risk is a very challenging task. One of the main
reasons is due to the diverse elements involved in the quantification process.
Although the use of Basic Indicator and the Standardised approach in measuring
operational risk exposures have been suggested by IFSB, however, both
approaches are somewhat inaccurate as the suggested methods are used to calculate
operational risk capital. In other words, it is a requirement to set aside a certain
amount of capital to cater for operational risk. Consequently, it will not come as a
surprise that such an approach will result in a high or low of economic capital
number; hence, operational risk capital is over estimated or under estimated.
Nonetheless, a very essential step which is actually overlooked in the process;

that is the measurement of operational risks itself. In this respect, this paper
proposes an approach for the measurement of operational risk, namely Delta
Gamma Sensitivity Analysis-Extreme Value Theory. This model is an integrated
measurement method which caters for two types of operational risks, namely high
frequency-low severity (HF-LS) and low frequency-high severity (LF-HS) risks.
The strength of the model lies in its accuracy in measuring the causality taking
place in the value adding process in Islamic banking operations. Moreover, an
elasticity based sensitivity analysis employed in the first leg of the model would be
a better alternative to the common partial derivatives based sensitivity analysis
since it would not neglect the relative parameter changes that occur in the causality
models.
The proposed DGSA-EVT model would also give a number advantage to the
operational risk managers; first, it is a reflection of potential loss that is not merely
based on actual loss figure which is rarely available. This aspect is very crucial
since in most cases, operational losses are not recorded, especially in an Islamic
bank. To the best of the authors knowledge, there is not any single Islamic bank
which discloses publicly the magnitude of its operational losses. Second, the
models reflect the quality of the operations in the banks. Thus, it can be perceived
that a bank with a better model is likely to have more effective operational risk
management. Third, since the error rates are relative errors based on exposures, the
models are related to the size of the firms business.
Nevertheless, the paper is theoretical and analytical in nature. Testing the
proposed model, therefore, is needed to assess the workability of the model.
Indeed, it is a quite daunting task considering that data availability can be a
hindrance to such a test.
Bibliography
Accounting and Auditing Organisation for Islamic Financial Institutions. (2005).
Shar ah Standards. Manama: AAOIFI.
Arzac, E., Bawa, V. (1977). Portfolio Choice and Equilibrium in Capital Markets
with Safety First Investors, Journal of Financial Economics, 4, pp. 277-288.
Basel Committee on Banking Supervision. (2006). International Convergence of
Capital Measurement and Capital Standards: A Revised Framework
Basel Committe on Banking Supervision Standard. (2010). BCBS on Operational
Risk-Supervisory Guidelines for the Advanced Measurement Approach
Bocker, K. and C. Kluppelberg. (2005). Operational VAR: A Closed-Form
Approximation, Risk (December), pp. 9093.
Borgonovo, E. and Apostolakis, G.E., (2001a). A New Importance Measure for
Risk Informed Decision-Making, Reliability Engineering and System Safety,
72 (2), pp. 193-212.
Borgonovo, E. and Apostolakis, G.E., (2001b). On Local Sensitivity Measures, in
Proceedings of Sensitivity Analysis of Model Output, Madrid, Spain.
Borgonovo, E. and Peccati, L. (2004). Sensitivity Analysis in Investment Project
Evaluation, International Journal of Production Economics, 90, pp. 17-25.
Chavez,-Demoulin, V. Embrechts, Pl. and Neslehova, J. (2006). Quantitative
Models for Operational Risk: Extremes, Dependence and Aggregation, Journal
of Banking and Finance, 30, pp. 2635-2658
Cheok, M.C., Parry, G.W. and Sherry, R.R. (1998). Use of Importance Measures
in Risk Informed Regulatory Applications, Reliability Engineering and System
Safety, 60, pp. 213-226.
Chernobai, Anna S., Rachev, Svetlozar T., and Fabozzi, Frank J. (2007).
Operational Risk: A Guide to Basel II Capital Requirement, Models, and
Analysis. Hoboken, New Jersey: John Wiley & Sons, Inc.
Coleman, R. (2003). Operational Risk Modelling for Extremes, Operational Risk
(January), pp. 6-9.
Consiglio, A. and S. A. Zenios. (2003). Model Error in Enterprise-wide Risk

Management: Insurance Policies with Gurantees. in Advances in Operational
Risk: Firm-wide Issues for Financial Institutions (2nd edition), London: Risk
Books.
Damodaran, Aswath. (2001). Corporate Finance: Theory and Practice. New York:
John Wiley & Sons, Inc.
De Fontnouvelle, Patrick., De-Jesus-Rueff, Virginia., Jordan, John., and
Rosengren, Eric. (2006). Capital and Risk: New Evidence on Implications of
Large Operational Losses, Journal of Money, Credit, and Banking, 38, , pp.
1819-46
De Fontnouvelle, Patrick., Jordan, John., and Rosengren, Eric. (2004). Implications
of Alternative Operational Risk Modelling Techniques, Working Paper, Federal
Reserve Bank of Boston.
Drudi, F., Generale, A. and Majnoni, G. (1997). Sensitivity of VaR Measures to
Different Risk Models, Temi di discussione no. 317, Banca dItalia.
Dutta, K. and J. Perry. (2007). A Tale of Tails: An Empirical Analysis of Loss
Distribution Models for Estimating Operational Risk Capital. FRB of Boston
Working Paper No. 06-13
Ebnother, S., Vanini, P., McNeil, A., Antonilez-Fehr, P. (2001). Modelling
Operational Risk. Working Paper, ETH Zurich.
Froot, K., and Stein, J. (1998). Risk Management, Capital Budgeting and Capital
Structure Policy for Financial Institution, Journal of Financial Economics, 47,
pp. 55-82.
Fujii, K. (2005). Building Scenarios. in E. Davis (ed.), Operational Risk:
Practical Approaches to Implementation. London: Risk Books.
Giraud, J. R. (2005). Managing Hedge Funds Exposure to Operational Risks, in
E. Davis (ed.), Operational Risk: Practical Approaches to Implementation.
London: Risk Books.
Gollier, C., Koehl, P.F., Rochet, J.C. (1996). Risk Taking Behaviour with Limited
Liability and Risk Aversion. DP 96-13, Wharton Financial Institutions Centre.
Gourieroux, C., Laurent, J.P. and Scaillet, O. (2000). Sensitivity Analysis of

Values at Risk, Journal of Empirical Finance, 7, pp. 225-245.
Hassan, M. Kabir and Dicle, Mehmet F. (2005). Basel II and Capital Requirements
for Islamic Banks. Paper presented at the 6th International Conference at Islamic
Economics, Banking and Finance, organised by Islamic Research and Training
Institute-Islamic Development Bank, Bank Indonesia, and Ministry of Finance
Republic of Indonesia.
Holmes, M. (2003). Measuring Operational Risk: A Reality Check, Risk (16), pp.
8487.
Islamic Financial Services Board. (2005). Capital Adequacy Standard for
Institutions (Other than Insurance Institutions) Offering Only Islamic Financial
Services. Available at <http://www.ifsb.org> Access date 11th December 2008
Ismail b., Abdul Ghafar and Ahmad Azam b. Sulaiman. (2005). Cyclical Patterns
in Profits, Provisioning and Lending of Islamic Banks and Procyclicality of the
New Basel Capital Requirement. Paper presented at the Sixth International
Conference on Islamic Economics and Finance: Islamic Economics and
Banking in the 21st Century, organized by Bank Indonesia, Islamic Research
and Training Institute, The International Association for Islamic Economics,
and University of Indonesia.
Izhar, Hylmun. (2010). Identifying Operational Risk Exposures in Islamic
Banking, Kyoto Bulletin of Islamic Are Studies, 3(2), pp. 17-53.
Jackson-Moore, Elisabeth. (2007). Measuring Operational Risk, in Simon Archer
and R.A.A.Karim (eds.), Islamic Finance: The Regulatory Challenge.
Chichester: John Wiley & Sons (Asia) Pte Ltd.
Jang, J. (2004). The Laplace Transform of the Distribution of the Cox Process with
Shot Noise Intensity and Its Application to Reinsurance and Operational Risk.
Working Paper, University of New South Wales.
Kahf, Monzer. (2004). Basel II: Implications for Islamic Banking. Paper presented
at the Sixth International Conference on Islamic Economics and Finance:
Islamic Economics and Banking in the 21st Century, organized by Bank
Indonesia, Islamic Research and Training Institute, The International
Association for Islamic Economics, and University of Indonesia.
Khan, Tariqullah and Ahmed, Habib. (2001). Risk Management: An Analysis of
Issues in Islamic Financial Industry. Occasional Paper No. 5. Jeddah: Islamic

Research and Training Institute, Islamic Development Bank.
Kimball, R. (1997). Innovations in Performance Measurement in Banking, New
England Economic (May-June), pp. 23-38.
King, Jack L. King (2001). Operational Risk: Measurement and Modelling.
Chichester: John Wiley & Sons, Ltd.
Levy, H., Sarnat, M. (1972). Safety first: An Expected Utility Principle. Journal
of Financial and Quantitative Analysis, 7, pp. 1829-1834.
Lindskog, F., McNeil, A. (2003). Common Poisson Shock Models: Applications
to Insurance and Credit Risk Modelling. ASTIN Bulletin, 33 (2), pp. 209-238.
Manganelli, S. (2004). Asset Allocation by Variance Sensitivity Analysis,
Journal of Financial Econometrics, 2(3), pp. 370-389.
McNeil, A.J., and Frey, R. (2000). Estimation of Tail Related Risk Measures for
Heteroskedastic Financial Time Series: An Extreme Value Approach, Journal
of Empirical Finance, 7(3-4), pp. 271-300.
Moosa, Imad A. (2007). Operational Risk Management. New York: Palgrave
MacMillan
Moscadelli, M. (2004). The Modelling of Operational Risk: Experience with the
Analysis of the Data Collected by the Basel Committee. Technical Report No.
517, Banca dItalia.
Muljawan, Dadang. (2005). A Design for Islamic Banking Rating System: An
Integrated Approach. Paper presented at the Sixth International Conference on
Islamic Economics and Finance: Islamic Economics and Banking in the 21st
Century, organized by Bank Indonesia, Islamic Research and Training Institute,
The International Association for Islamic Economics, and University of
Indonesia.
Peccia, A. (2003). Using Operational Risk Models to Manage Operational Risk.
Pp. 363384 in Operational Risk: Regulation, Analysis and Management, ed. C.
Alexander. London: Prentice Hall-Financial Times.
Porter, M. E. (1985). Competitive Advantage: Creating and Sustaining Superior
Performance. New York: The Free Press.
Roy, A. (1952). Safety First and the Holding of Assets. Econometrica, 20, pp.
431-449.
Saltelli, A. (1997). The Role of Sensitivity Analysis in the Corroboration of
Models an its Link to Model Structural and Parametric Uncertainty, Reliability
Engineering and System Safety, 57, pp. 1-4.
Saltelli, A. (1999). Sensitivity Analysis: Could Better Methods be Used?,
Journal of Geophysical Research, 104, pp. 3789-3793.
Saltelli, A., Tarantola, S. and Chan, KPS. (1999). A Quantitative ModelIndependent Method for Global Sensitivity Analysis of Model Output,
Technometrics, 41 (1), pp. 39-56.
Santomero, A., Babbel, D. (1996). Risk Management by Insurers: An Analysis of
the Process. DP 96-16, Wharton Financial Institutions Centre.
Simon, C.P., and Blume, L. (1994). Mathematics for Economists. New York:
W.W. Norton & Co.
Smithson, Charles, and Song, Paul. (2004). Quantifying Operational Risk, Risk
(July), pp. 50-52.
Stoughton, N., and Zechner, J. (1999). Optimal Capital Allocation Using RAROC
and EVA, DP UC Irvine.
Sundararajan, V. (2005). Risk Measurement and Disclosure in Islamic Finance and
the Implications of Profit Sharing Investment Accounts. Paper presented at the
Sixth International Conference on Islamic Economics and Finance: Islamic
Economics and Banking in the 21st Century, organized by Bank Indonesia,
Islamic Research and Training Institute, The International Association for
Islamic Economics, and University of Indonesia.
Turanyi, T., and Rabitz, H. (2000). Local Methods, in A. Saltelli, K. Chan and
E.M. Scott (eds.), Sensitivity Analysis, Chichester: John Wiley & Sons.
APPENDIX43
The appendix discusses different dimensions of proposed operational risk categories
(ORCs) in different types of Islamic financial contracts. As can be seen in the table below,
the five dimensions of operational risk categories are Shar ah compliance risk (SR),
fiduciary risk (FR), people risk (PR), legal risk (LR), and technology risk (TR). The first
three dimensions are, by nature, internally inflicted; while the fourth one is naturally from
external source. As for technology risk (TR); it can originate from either internal or external
operational failures.
The Dimensions of Operational Risk Categories (ORCs) in

Islamic Financial Contracts
Contracts
Mur ba ah
Salam
Isti n
43
Internal Risks
Shar ah
Compliance Risk
(SR)
Exchange of
money and
commodity
needs to be
ensured
In the event
of late payment,
penalty must be
avoided as it will
tantamount to
rib .
Final
payment of
monetary
rewards must be
concluded in
advance
Penalty
clause is
illegitimate in
the event of
sellers default
in delivering the
goods
In parallel
salam, execution
of second salam
contract is not
contingent on
the settlement of
the first salam
contract
Should not
be used as a
Fiduciary Risk
(FR)
People Risk
(PR)
Inability to
meet the
specified
product
stipulated in the
contract
External Risks
Technology
Risk (TR)
Legal Risk
(LR)
Technology
Risk (TR)
Fail to
deliver the
product
Incompatibility
of the new
accounting
software
Products to be
sold must be
legally owned by
the bank
System
failures and
external
security
breaches
Inability to
meet the
specified
product
stipulated in
the contract.
Delivery
of inferior
goods cannot
be accepted
Mismatch in
the
commoditys
specification
due to
inability of
seller to
provide the
exact
product
mentioned in
the contract.
Incompatibility
of the new
accounting
software
Goods must be
delivered when it
is due, as agreed
in the contract
Specification
mismatching
in
commodities
productions
agreed in the
contract
Need to ensure
the quality
Inability to
deliver the
Incompatibility
of the new
This section is heavily drawn from Izhar (2010)
Disagreement
with the sub-
Specification
mismatching
Ij rah
Mush rakah
Mu
rabah
legal device; e.g.

the party
ordering the
product to be
produced is the
manufacturer
himself
In parallel
isti n ,
contracts should
be separated to
avoid two sales
in one deal
Need to
ensure that
leased asset is
used in a
Shar ah
compliant
manner
In ij rah
muntahia
bittamleek, an
option to
purchase cannot
be enforced.
Profit allocation is
based on actual
profit, not
expected profit
standards of the
products
product on
time
accounting
software
contractor or the
customer in the
event of
remedying the
defects
in
commodities
productions
agreed in the
contract
Major
maintenance of
the leased asset
is the
responsibility
of the banks or
any party acting
as lessor.
Lessor needs
to
understand
that in the
event of
payment
delay, rental
due cannot
be increased
as clearly
exemplified
by AAOIFI
Incompatibility
of the new
accounting
software
Enforcement of
contractual right
to repossess the
asset in case of
default or
misconduct by
the lessee
Losses of
information
on the leased
assets
specified in
the contract
due to
external
security
breaches
Inadequate
monitoring of
the financial
performance of
the venture
Lack of
technical
expertise in
assessing the
project
Incompatibility
of the new
accounting
software
A mixture of
shares in one
entity may lead
to legal risk if
the regulation
does not
facilitate such
action
Losses of
information
on the
projects
specified in
the contract
due to
external
security
breaches
Profit allocation is
based on actual
profit, not
expected profit
Inadequate
monitoring of
the business
Inability to
provide
regular and
transparent
financial
performance
of the
project
Incompatibility
of the new
accounting
software
Misinterpretation
of civil law upon
implementation
of Shar ah
compliant
mu rabah
Losses of
information
on the
projects
specified in
the contract
due to
external
security
breaches
Mur ba ah
Mur ba ah is selling a commodity as per the purchasing price with a defined
and agreed profit mark-up44. This mark-up may be a percentage of the selling
price or a lump sum. Moreover, according to AAOIFI standard, this transaction
may be concluded either without a prior promise to buy, in which case it is called
44
Accounting and Auditing Organisation for Islamic Financial Institutions (AAOIFI) on Shariah
Standards, 2005.
ordinary mur ba ah, or with a prior promise to buy submitted by a person

interested in acquiring goods through the institution, in which it is called a
banking mur ba ah, i.e. mur ba ah to the purchase orderer. This transaction is
one of the trust-based contracts that depends on transparency as to the actual
purchasing price or cost price in addition to common expenses.
Mur ba ah is the most popular contract in terms of its use, since most of
Islamic commercial banks operating worldwide rely on this contract in generating
income. Different dimensions of operational risk which can arise in mur ba ah
transaction are as follows:
Shar ah compliance risk (SR); may arise if the Islamic banks give money,
instead of commodity, which will then result in the exchange of money and
money. This is prohibited in Shar ah, since the exchange of money with
money, plus additional amount above the principal and paid in different time
will tantamount to rib . AAOIFI Shar ah standard also requires Islamic
banks to own, legally, the commodity before they sell it to the customers. It
is important to note that the sequence of the contract is very central in
mur ba ah transaction. Inability or failure to conform with the sequence and
Shar ah requirement will result in the transaction to be deemed illegitimate.
Fiduciary risk (FR); this risk arises due to the inability to meet the specified
commodity stipulated in the contract.
People risk (PR); the risk can result from two sides, seller as well as buyer.
PR from the seller side occurs if Islamic banks fail to deliver the specified
product agreed in the contract on due date, while PR from the buyer side
takes place when the buyers does not keep their promise to buy the
commodity. This can happen in the binding mur ba ah contract.
Legal risk (LR); profit originated from mur ba ah cannot be equated with
interest, although it looks similar. The main difference is because the
resulting profit is tied with the underlying commodity. This might create
legal problem as in certain countries, the regulators only give limitation on
interest rate, not profit rate. Hence, the absence of so called profit rate cap
has the potential to crate legal problems if there is any dispute. Another
potential problem can occur at the contract signing stage, since the contract
requires the Islamic bank to purchase the asset first before selling it to the
customer; the bank needs to ensure that the legal implications of the contract
properly match the commercial intent of the transactions.
Technology risk (TR); may result from an incompatibility of the new
accounting software or an external system failure.
Salam and Parallel Salam

AAOIFI Shar ah standards define salam as a transaction of the purchase of a
commodity for the deferred delivery in exchange for immediate payment. It is a
type of sale in which the price, known as the salam capital, is paid at the time of
contracting while the delivery of the item to be sold, know as al-muslam fihi (the
subject matter of a salam contract), is deferred. The seller and the buyer are known
as al-muslam ilaihi and al-muslam or rabb al-salam respectively. Salam is also
known as salaf. Parallel salam occurs when the seller enters into another separate
salam contract with a third party to acquire goods, the specification of which
corresponds to that of the commodity specified in the first salam contract.
Shar ah compliance risk (SR); one of the very central conditions in salam
contract is that the payment of salam capital must be paid full in advance. If
payment is delayed, the transaction is not called salam45.Any delay in
payment of the capital and dispersal of the parties renders the transaction a
sale of debt for debt, which is prohibited, and the scholars agreed on its
prohibition. Another aspect, which might lead to SR may also occur in
parallel salam; this will take place if the execution of the second salam
contract is contingent on the execution of the first salam contract. Penalty
clause is also not allowed, in the event of a sellers default in delivering the
good. The basis for not allowing penalty in salam is because al-muslam fihi
(the subject matter of a salam contract) is considered to be a debt; hence it is
not permitted to stipulate payment in excess of the principal amounts of
debt46.
Fiduciary risk (FR); salam is generally associated with the agricultural
sector. The buyer must either rejects goods of an inferior quality to that
specified in the contract, or accept them at the original price. In the latter
case, the goods would have to be sold at a discount (unless the customer
under a parallel salam agreed to accept the goods at the originally agreed
price).
People risk (PR); can arise due to a sellers default in delivering the
commodity or due to the commoditys specification mismatching. Financial
institutions may minimise such type of operational risks by asking from the
seller guarantees that they are following a quality management system or
following any standard system, or by asking for references on past promises
on salam contract or by collateralising their losses via insurance policies.
45
46
AAOIFI, 2005, p. 172

AAOIFI, 2005, p. 173
Legal risk (LR); Islamic banks may face legal risk if the goods cannot be
delivered at the specified time (unless the customer under parallel salam
agrees to modify the delivery date).
accounting software or the system fails to specify precisely the commodities
agreed in the contract.
Isti n
and Parallel Isti n
Isti n is another type of forward contract, but the role of an Islamic bank as a
financial intermediary differs from that in a salam contract. In this case, the bank
contracts to supply a constructed asset (such as a building or a ship) for a customer.
In turn, the bank enters into a parallel isti n with a sub-contractor in order to have
the asset constructed. Its reliance on the parallel isti n counterparty (the subcontractor) exposes it to various operational risks, which need to be managed by a
combination of legal precautions, due diligence in choosing sub-contractors, and
technical management by appropriately qualified staff or consultants of the
execution of the contract by the sub-contractor. Islamic banks that specialise in
isti n financing may have an engineering department. Risks may include the
following:
Shar ah compliance risk (SR); could arise if Isti n is being used as a legal
device for mere interest based financing. For instance, an institution buys
items from the contractor on a cash payment basis and sells them back to the
manufacturer on a deferred payment basis at a higher price; or where the
party ordering the subject matter to be produced is the manufacturer himself;
or where one third or more of the facility in which the subject matter will be
produced belongs to the customer. All the circumstances mentioned above
would make the deal an interest based financing deal in which the subject
matter never genuinely changes hands, even if the deal won through
competitive bidding. This rule is intended to avoid sale and buy back
transactions (bay al- nah). In parallel isti n , the separation of contracts is
a must, hence this is not an instance of two sales in one deal, which is
prohibited.
Fiduciary risk (FR); the sub-contractor may fail to meet quality standards or
other requirements of the specification, as agreed with the costumer under
the isti n contract.
People Risk (PR); this may arise if the Islamic bank may be unable to deliver
the asset on time, owing to time overruns by the sub-contractor under the
parallel isti n , and may thus face penalties for late completion.
Legal risk (LR); Islamic banks may face legal risk if no agreement is reached
with the sub-contractor and the customer either for remedying the defects or
for reducing the contract price.
accounting software or the system fails to specify precisely the commodities
that would be produced in the contract.
Ij rah and Ij rah Muntahia Bittamleek
In simple terms, an ij rah contract is an operating lease, whereas ij rah
muntahia bittamleek is a lease to purchase. While operational risk exposures during
the purchase and holding of the assets may be similar to those in case of
mur ba ah, other operational risk aspects include the following:
Shar ah compliance risk (SR); the Islamic banks need to ensure that the
asset will be used in a Shar ah compliant manner. Otherwise, it is exposed
to non-recognition of the lease income as non-permissible.
Fiduciary risk (FR); major maintenance is the responsibility of an Islamic
bank as a lessor, as directed by AAOIFI Shar ah standards47. In addition to
that, it is the duty of the lessor to ensure that the usufruct is intact, and this is
not possible unless the asset is maintained and kept safe so that the lessor
may be entitled to the rentals in consideration for the usufruct. Thus,
deficiencies in maintaining such responsibility can be deemed to be sources
of FR in ij rah contract.
People risk (PR); lessor is not allowed to increase the rental due, in case of
delay of payment by the lesse, this is what AAOIFI clearly exemplifies.
Misunderstanding of this principle by the staff is a source of losses caused
by PR, because the income generated from this, is not permissible from
Shar ah point of view.
Legal risk (LR);the Islamic bank may be exposed to legal risk in respect of
the enforcement of its contractual right to repossess the asset in case of
default or misconduct by the lessee. This may be the case particularly when
the asset is a house or apartment that is the lessees home, and the lessee
enjoys protection as a tenant.
47
AAOIFI, 2005, p. 154
Technology risk (TR); may occur due to an incompatibility of the new

accounting software or losses of information on the leased assets due to
external security breaches.
Mush rakah
Mush rakah is a profit and loss sharing partnership contract. The Islamic bank
may enter into a mush rakah with a customer for the purpose of providing a
Shar ah compliant financing facility to the customer on a profit and loss sharing
basis. The customer will normally be the managing partner in the venture, but the
bank may participate in the management and thus be able to monitor the use of the
funds more closely. Typically, a diminishing mush rakah will be used for this
purpose, and the customer will progressively purchase the banks share of the
venture. Operational risks that may be associated with mush rakah investments are
as follows:
Shar ah compliance risk (SR); the source of SR may arise due to the final
allocation of profit taking place based on expected profit. AAOIFI
commands that it is necessary that the allocation of profit is done on the
basis of actual profit earned through actual or constructive valuation of the
sold assets48.
Fiduciary risk (FR); any misconduct or negligence of the partners are the
sources of FR. This can happen in the absence of adequate monitoring of the
financial performance of the venture.
People risk (PR); lack of appropriate technical expertise can be a cause of
failure in a new business activity.
Legal risk (LR); an Islamic bank which enters into mush rakah contract
needs to acquire some shares from separate legal entity that undertake
Shar ah compliant activities. A mixture of shares in one entity may lead to
legal risk if the regulation does not allow doing such action.
Technology risk (TR); may occur due to an incompatibility of the new
accounting software or losses of the precise information on projects
undertaken due to external security breaches.
Mu rabah
Mu rabah is a profit sharing and loss bearing contract under which the
financier (rabb al m l) entrusts his funds to an entrepreneur (mu rib). The
48
AAOIFI, 2005, p. 205
exposure of operational risk in mu rabah is somewhat similar to that of

mush rakah. However, since this type of contract may be used on the assets side of
the balance sheet, as well as being used on the funding side for mobilising
investment accounts, the operational risk is first analysed from the assets-side
perspective and then from the funding side perspective (which is related to
fiduciary risk)
Asset-side Mu rabah
Contractually, an Islamic bank has no control over the management of the
business financed through this mode, the entrepreneur having complete freedom to
run the enterprise according to his best judge judgement. The bank is contractually
entitled only to share with the entrepreneur the profits generated by the venture
according to the contractually agreed profit sharing ratio. The entrepreneur as
mu rib does not share in any losses which are borne entirely by the rabb al m l.
The mu rib has an obligation to act in a fiduciary capacity as the manager of the
banks funds, but the situation gives rise to moral hazard especially if there is
information asymmetrythat is, the bank does not receive regular and reliable
financial reports on the performance of the mu rib. Hence, in addition to due
diligence before advancing the funds, the bank needs to take precautions against
problems of information asymmetry during the period of investment.
Funding-side Mu rabah
Profit-sharing (and loss bearing) investment accounts are a Shar ah compliant
alternative to conventional interest-bearing deposit account. Since a mu rabah
contract is employed between the Islamic bank and its investment account holders,
the investment account holders (IAHs) share the profits and bear all losses without
having any control or rights of governance over the Islamic bank. In return, the
Islamic bank has fiduciary responsibilities in managing the IAHs funds. The IAHs
typically expect returns on their funds that are comparable to the returns paid by
competitors (both other Islamic banks and conventional institutions), but they also
expect the Islamic bank to comply with Shar ah rules and principles at all times. If
the Islamic bank is seen to be deficient in its Shar ah compliance, it is exposed to
the risk of IAHs withdrawing their funds and, in serious cases, of being accused of
misconduct and negligence. In the latter case, the funds of the IAHs may be
considered to be a liability of the Islamic bank, thus jeopardising its solvency.

Measuring or Exposures

Загружено:

Сведения о документе

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Measuring or Exposures

Загружено:

Авторское право:

Доступные форматы

Islamic Economic Studies

Vol. 20, No.1, June 2012

Measuring Operational Risk Exposures in Islamic Banking:

See Moosa, 2007

46 Islamic Economic Studies, Vol. 20 No. 1

Hylmun Izhar: Measuring Operational Risk Exposures in Islamic Banking 47

(BIA) and the Standardised Approach (TSA) as methods to calculate operational

48 Islamic Economic Studies, Vol. 20 No. 1

of operational risk. Surely, a constantly changing financial environment has made

Source: Smithson and Song (2004)

Hylmun Izhar: Measuring Operational Risk Exposures in Islamic Banking 49

2.2. Process approach

50 Islamic Economic Studies, Vol. 20 No. 1

2.3. Factor Approach

( OperationalRisk )i ( Factor1 ) ( Factor 2 )

Hylmun Izhar: Measuring Operational Risk Exposures in Islamic Banking 51

loss distribution or a distribution for the frequency of occurrence of losses and a

Islamic Financial Services Board, op cit. p. 22.

52 Islamic Economic Studies, Vol. 20 No. 1

measurementbased on externally provided rating categoriesand within this

PSIA refers to profit sharing investment account

Hylmun Izhar: Measuring Operational Risk Exposures in Islamic Banking 53

54 Islamic Economic Studies, Vol. 20 No. 1

Hylmun Izhar: Measuring Operational Risk Exposures in Islamic Banking 55

medium or long term, is a hybrid of shirka (partnership) and mu rabah,

56 Islamic Economic Studies, Vol. 20 No. 1

Identification of Risk Factors in

Establish earnings functions related to risk factors in each IGC. How?

1st step (static process): Given

2st step (dynamic process):

Value at Risk for

Decide maximum OLD as

Hylmun Izhar: Measuring Operational Risk Exposures in Islamic Banking 57

Establish the business model with income generating channel

4.2. Key Features of DGSA

58 Islamic Economic Studies, Vol. 20 No. 1

u( E ) fu( X 1 )...u( X n )) ( Lunassignable Lunassignable )

Earnings are described as a function of a set of causal factors. For example,

Hylmun Izhar: Measuring Operational Risk Exposures in Islamic Banking 59

Whereby x is the mean of the analysed operational risk variable.

60 Islamic Economic Studies, Vol. 20 No. 1

Then sensitivity is defined as

The method of combining measurement uncertainties from various factors and

Hylmun Izhar: Measuring Operational Risk Exposures in Islamic Banking 61

It is expected that partnership type of financing, such as mu rabah and

62 Islamic Economic Studies, Vol. 20 No. 1

Therefore, using Elasticity (E) is considered to be a better alternative as

D can be interpreted as the ratio of the (infinitesimal) change in Y caused by a

Hylmun Izhar: Measuring Operational Risk Exposures in Islamic Banking 63

for example, the sum of the Di (i=1,,n) of all parameters is equal to

In the hypothesis of uniform parameter changes (H1) (dxj=dxs j,s), the

Hypothesis of proportional changes (H2) dx j j , would result in:

It can be shown that D generalises other local SA techniques as the Fussel-Vesely

E. Borgonovo and L. Peccatti, op cit., p. 23

64 Islamic Economic Studies, Vol. 20 No. 1

5. Determination of Risk Factor Contribution

Hylmun Izhar: Measuring Operational Risk Exposures in Islamic Banking 65

V. Chavez-Demoulin, P. Embrechts, J. Neslehova, Quantitative Models for Operational Risk :

66 Islamic Economic Studies, Vol. 20 No. 1

Severity [ loss amount>threshold ()]

Maximum likelihood method

Fitted to GPD (Generalised Pareto Distribution)

Fitted to Poisson Distribution

Combined in Monte Carlo Simulation